Path: ...!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail From: Woozy Song Newsgroups: comp.lang.fortran Subject: Intel oneAPI compiler not happy with AVX512 Date: Fri, 12 Apr 2024 17:58:38 +0800 Organization: A noiseless patient Spider Lines: 9 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Injection-Date: Fri, 12 Apr 2024 11:58:42 +0200 (CEST) Injection-Info: dont-email.me; posting-host="2e3cd12f0db2abf4bb11a988e1f5fc73"; logging-data="2416045"; mail-complaints-to="abuse@eternal-september.org"; posting-account="U2FsdGVkX19oahbdmvaszWG5Zb+00NX4FTl/i7r9kI8=" User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101 Firefox/91.0 SeaMonkey/2.53.18.2 Cancel-Lock: sha1:GohNF3T7bSewRX59d5JAf43o5dU= X-Mozilla-News-Host: news://news.eternal-september.org:119 Bytes: 1605 I have a package that contains nearly 500 f90 files. I noticed that in the makefile, it was using -ax (automatic dispatch ) up to AVX2. I added COMMON-AVX512 to the options. One thing I noticed was that compilation is now slower. It seemed to stop for 30 seconds when it hit one particular subroutine. But main problem was that after a few hundred files, it aborted with unexpected internal error. The offending file wasn't doing anything exotic, mostly a linear regression. This is 2024.1 version of the compiler, pretty much the latest. Has anybody else compiled for AVX512?